**2. Materials and Methods**

To deposit coatings, a VIT-2 unit (VIT—IDTI RAS, Moscow, Russia) was used, which implemented the filtered cathodic vacuum arc deposition (FCVAD) technology [5,63–69].

Three cathode systems were used, in which cathodes of Al 99.1 at.%, Ti 99.9 at.%, and, depending on the type of the coating to be deposited, Cr–Mo (50:50 at.%) or Zr–Nb (50:50 at.%) were installed, respectively. Cathodes of cylindrical shape with a diameter of 80 mm were used.

The parameters for the process of coating deposition are presented in Table 1.


**Table 1.** Parameters of stages of the technological process of the deposition of coatings.

Note: *I*Ti = current of titanium cathode, *I*Zr–Nb = current of Zr–Nb cathode, *I*Al = current of Al cathode, *I*Cr–Mo = current of Cr–Mo cathode, *p*<sup>N</sup> = gas pressure in chamber, *U* = voltage on substrate. Table rotation frequency *n* = 0.7 rpm.

Due to the fact that the VIT-2 unit uses a micro-droplet reduction system, the droplets in the coating stemming from the Al-cathode are reduced by up to 98% [5,70] and the droplets in the coating stemming from the other cathodes are reduced by up to 85% [71–73]. Microdroplets in the structure of the coatings under study are rather rare. The sizes of the detected microdroplets based on Mo, Cr, Ti, and Zr–Nb usually do not exceed 3 μm (individual microdroplets up to 12 μm in size are rarely observed), because larger microdroplets are separated and do not enter a product (several sources consider the mean microdroplet size of 10–25 μm [1,2]). Samples were subjected to the microstructural studies using an SEM FEI Quanta 600 FEG (SEM 1, Waltham, MA, USA) and a Carl Zeiss EVO 50 with a Bruker energy-dispersive spectrometer (SEM 2, Billerica, MA, USA). A high-resolution transmission electron microscope (HR TEM) JEM 2100 (JEOL, Tokyo, Japan) at an accelerating voltage of 200 kV was used to investigate the coating nanostructure. The analysis of the elemental composition of the coatings was conducted at TEM using the EDX system INCA Energy (OXFORD Instruments, Abingdon, UK) in STEM (scanning transmission electron microscopy) mode. A Strata FIB 205 (FEI) was used to prepare TEM samples.

### **3. Results and Discussion**

The available results from studies of the microstructure of coatings of various compositions and architecture, show that the detected microdroplets can be classified by several parameters (for example, by the shape of a microdroplet and its position in the coating structure) [13,74–76]. Accordingly, the influence of various microdroplet types on the structure of the coating and its fracture pattern also differ. In particular, the stage of the coating deposition process, at which a microdroplet is embedded into the coating structure, has a significant effect. A microdroplet can be embedded at the initial stage of the deposition process, and in such case, it often directly adheres to the substrate surface. A microdroplet can be embedded into the coating structure during the formation of a sequence of nanolayers. Finally, a microdroplet can be formed at the finishing stage of the deposition process. In such case, it penetrates into the coating surface and, since the coating has not yet cooled down and stays considerably ductile, forms a crater around itself and thus, deforms the coating structure. It should be noted that it is very difficult to directly study the process of the formation of a microdroplet, its movement inside the chamber, and penetration into the coating structure because of the small sizes of the microdroplets, their high movement velocity (which varies in the range of 6–120 m/s [1] (according to other data, 100–1000 m/s, depending on the temperature and the evaporated material [16,17]) and the complicated monitoring in the conditions of a plasma flow. Thus, at present, there are only two available ways to study the dynamics of microdroplet development, i.e., mathematical modeling and the study of solidified microdroplets and their effect on the coating structure. Microdroplets can also be classified by their final shape, which may indicate their state at the moment of impact onto the coating. In particular, there are microdroplets of an almost ideal spherical shape, due to which it can be concluded that the microdroplets have been embedded into the coating structure in a solidified state, otherwise their noticeable deformation would be inevitable. At the same time, there are lens-shaped microdroplets. This shape is typical for a microdroplet hitting the surface in a molten (liquid) state. Finally, there are microdroplets of a transition shape, that is, with signs of noticeable deformation or drop-shaped, which may indicate that at the moment of its penetration into the coating, the microdroplet was no longer liquid, but still was considerably ductile. Let us consider in more detail the structure and shapes of microdroplets of various metals. Since microdroplets of Al are almost completely separated during the deposition process, they could not be detected in the structure of the coatings, despite the fact that Al is contained in the compositions of all the coatings under study.

The structure of a microdroplet embedded into the structure of the Ti–TiN–(Ti,Cr,Mo,Al)N coating was considered. The images reveal (Figure 1) the sections of several clearly pronounced areas with noticeably different crystal sizes, which can indicate the stages of the microdroplet structure formation during the cooling process. Their formation, presumably, occurred during the microdroplet movement from the cathode towards the deposition surface and during the penetration of the microdroplet into the coating structure. The microdroplet core is formed mainly by molybdenum and chromium (Figure 1c). Area I with grains of extremely small sizes (about 3–5 nm) was formed around the microdroplet core. Then areas II and III were formed, and the grain sizes in each new area were noticeably larger than in the previous one (of about 20–40 nm in area III). Presumably, all the three areas form around the core when a microdroplet is moving from the cathode towards the deposition surface, until it hits the surface. The cooling rate of the microdroplet substance is related to the thermal conductivity coefficient (see Table 2). Since aluminium Al and molybdenum Mo are characterized by the highest thermal conductivity, while titanium Ti and zirconium Zr by the lowest thermal conductivity, the probability of the solidification of a microdroplet before its contact with the surface is higher for microdroplets of Al and Mo than for microdroplets of Ti and Zr. The size of a microdroplet and the length of its trajectory from the cathode to the deposition surface are essential. In the microdroplet, the core is formed mainly by Mo and Cr (Figure 1d), while the microdroplet composition also includes Ti and Al. Since the composition of the Mo–Cr cathode does not include Ti and Al, their presence in the microdroplet can be explained by their deposition on the microdroplet surface during its movement from the cathode to the deposition surface.


**Table 2.** Coefficients of thermal conductivity of metals depending on temperature, W⁄(m × K) [77].

Let us consider the differences in the selected area electron diffraction (SAED) pattern for the coating structure (Figure 1e) and the microdroplet structure (Figure 1f). The analysis of the SAED pattern for the Ti–TiN–(Ti,Cr,Mo,Al)N coating (Figure 1e) finds the presence of a basic phase of (Cr,Ti,Mo,Al)N with the space group of Fm3m and a small amount of AlN (given the intensity of the rings) with the space group of P6.3mc. The SAED pattern for the coating demonstrates the broadening of the rings, related to the (Cr,Ti,Mo,Al)N phase. This broadening apparently arises due to the presence of polycrystalline grains formed by several types of nitrides with the same space group Fm3m, but with slightly different interplanar spacings for the general interference indices (Miller indices, HKL). Table 3 contains the values of interplanar spacings d (for the first six HKL indices) for the nitrides forming the Ti–TiN–(Ti,Cr,Mo,Al)N coating under study.

**Figure 1.** (**a**–**c**) structure and (**d**) chemical composition of the microdroplet in the Ti–TiN–(Ti,Cr,Mo,Al)N coating (TEM), selected area electron diffraction (SAED) pattern for area A (**e**) and area B (**f**). The microdroplet core is surrounded by area I (crystal sizes are 3–5 nm), then by area II (crystal sizes are 8–15 nm), and then by area III (crystal sizes are 20–40 nm).



Table 3 also contains the interplanar spacings calculated under the SAED pattern for area B of the microdroplet under the study (Figure 1f). The SAED pattern for the microdroplet (area B) exhibits two differences compared to the SAED pattern for area A of the coating. Firstly, the SAED pattern for the microdroplet includes no AlN phase reflections, and, secondly, there is no noticeable broadening of the SAED rings. No broadening indicates that ether a certain nitride phase dominates in amount, or there are several nitride phases with very close values of d. Table 3 demonstrates that the calculated interplanar spacings correspond most closely to Mo2N (Fm3m). During the comparison of the elemental compositions of the coating and of the microdroplet, it can be noticed that in the region of the microdroplet, the amount of titanium decreases, and the amount of molybdenum noticeably increases, while the amount of chromium remains approximately the same. Based on the data obtained during the study of the microdroplet (investigation of its interplanar spacings and elemental composition), it can be assumed that the two phases—chromium nitride and molybdenum nitride—with considerably close interplanar spacings most likely appear in this region.

Thus, the microdroplet consists of chromium and nitrogen nitrides and not of pure metals. The above may confirm the hypothesis about the formation of a hard shell of a microdroplet due to the formation of nitrides that are more refractory than pure metal [34]. However, in this case, not only the shell of the microdroplet, but almost the entire microdroplet consists of nitrides, since the SAED pattern exhibits no rings associated with Cr or Mo.

The study of other examples showing how microdroplets of nearly spherical shape penetrate into the coating structure (Figure 2) reveals certain consistent patterns. Like in the case considered earlier, there is a microdroplet core with a fairly regular spherical shape, and the secondary structures, clearly distinguishable in Figure 2a,b, are formed around the microdroplet core. It should be noted that such a secondary structure is being formed around the microdroplet core not only on the side facing the cathode, but also on the opposite side, which is also noticeable in the earlier considered example (Figure 1). The presence of such structure formed around the spherical core of the microdroplet is difficult to explain if it is assumed that the structure was formed after the microdroplet had hit the deposition surface. In such case, no secondary structure would be formed on the microdroplet side opposite to the source. However, Figure 2b exhibits the microdroplet of Nb surrounded by the secondary structure, consisting of six layers of (Ti,Al,Nb,Zr)N, and the thickness of these layers noticeably increases on the side facing the source (with a thickness of about 250 nm). Meanwhile, such layers are also formed on the reverse side of the microdroplet (with a thickness of about 25 nm). The deposition of the secondary structure layers on the reverse side of the microdroplet core is possible only in cases when a microdroplet did not hit the deposition surface, that is, during the movement of the microdroplet from the cathode to the deposition surface. However, all previously obtained data indicate that the velocity of a microdroplet is very high and that it passes the distance from the cathode to the deposition surface within fractions of a second [1,16,17]. It is clear that the microdroplet cannot solidify during the above time and form a regular sphere (especially taking into account the fact that the coating deposition takes place in a highly discharged medium), and all the more, several layers of the secondary structure cannot form on its surface. Given that the turntable rotation frequency was 0.7 rev/min, it takes 42 s to deposit one layer and 252 s, that is, over 4 min, to deposit six layers. Based on the available methods and models, it can be assumed that free movement of the microdroplet core during such a long time is not possible. However, it should be noted that a moving microdroplet is affected by multidirectional plasma flows and significant magnetic fields. While there are three sources of plasma located at an angle of 90◦ to each other, a microdroplet is simultaneously affected by three plasma flows, and as it approaches the chamber center, the effect of all three flows can be balanced and at a certain point, such equilibrium effect can keep the microdroplet in a free state (that is, with no contact with the deposition surface or chamber walls). Another source of influence on the microdroplet, is the powerful magnetic field generated by the separation system. The combined effect of the above factors can lead to an effect in which the microdroplet can stay in a free state for a

relatively long time, sufficient for its cooling and deposition of a layered secondary structure on the microdroplet core.

**Figure 2.** (**a**) the microdroplet of Cr–Mo, embedded into the structure of the Ti–TiN–(Ti,Cr,Mo,Al)N coating, (**b**,**c**) the microdroplets of Nb–Zr, embedded into the structure of the Ti–TiN–(Ti,Al,Nb,Zr)N coating (SEM 2).

Given the small size and weight of the microdroplet core in the cases under study (with microdroplet diameters from 100 to 1000 nm), the above suggests the possibility of its long persistence in a free state. Of course, this hypothesis requires further development, implying thorough mathematical modeling of the process. It should be taken into account that in order to build an adequate model for the movement of a microdroplet, it will be necessary to consider the influence of multidirectional plasma flows (given the presence of at least three plasma sources operating simultaneously, as well as the possibility for the formation of vortex-like plasma flows). The influence of the powerful magnetic field of the plasma flow separation system should also be taken into account, as well as the effect of the internal discharged medium of the chamber, which makes the development of such a model rather complicated.

When a microdroplet has a larger core (3–5 μm), its weight also increases significantly. Accordingly, such microdroplets reach the deposition surface relatively quickly, while there is not enough time for the formation of the secondary structures or an insignificant amount of them are formed only on the side facing the plasma source (Figure 3). However, it is clear that such microdroplets still have enough time to cool down and solidify, although they stay considerably ductile. When microdroplets hit the deposition surface, they undergo plastic deformation, and they become ellipse-shaped (Figure 3b,c) or tear-shaped (Figure 3a). After the penetration of microdroplets into the coating structure, their deposited nanolayers are distorted.

**Figure 3.** Examples of the influence of microdroplets on the structure of the (**a**,**b**) Ti–TiN–(Ti,Cr,Mo,Al)N and (**c**) Ti–TiN–(Ti,Nb,Zr,Al)N coatings (SEM 1).

Microdroplets of even larger sizes (5–7 μm) also do not have enough time to form a secondary structure in the process of movement. When hitting the coating surface, the above microdroplets can form a crater and noticeably distort the nanolayer structure of the coating (Figure 4). It should be taken into account that immediately after the deposition, during the process of cooling, the coating layers retains their high ductility and when a considerably large object hits the freshly deposited coating layers, it can deform their structure (see Figure 4a, in this image, the very microdroplet was lost, possibly during the manufacturing of a section, but the trace of the microdroplet is clearly visible).

**Figure 4.** Examples of the influence of microdroplets on the structure of (**a**) Ti–TiN–(Ti,Cr,Mo,Al)N and (**b**) Ti–TiN–(Ti,Nb,Zr,Al)N coatings (SEM 1).

Finally, there are also lens-shaped microdroplets embedded into the coating structure, and their shape can be called "correct", because it is fully consistent with the existing models [14–19]. Although most of the microdroplets of the above shape have considerably large sizes and, accordingly, weights (with a diameter of 7–10 μm and more, see Figure 5a,b), there are also microdroplets of smaller sizes, which took the shape of a lens when hitting the deposition surface (the microdroplets of Ti shown in Figure 5c,d). The microdroplets of such a shape have noticeably less effect on the

distortion of the nanolayer structure of the coating. It can be assumed that the formation of lens-shaped microdroplets occurs in conditions when the microdroplet weight is large enough and the initial energy of the microdroplet movement allows it to move at a considerably high velocity. There are also cases when the interaction of plasma flows and magnetic fields does not have a significant influence on the velocity of the microdroplet movement or even contributes to the acceleration of such movement. A low thermal conductivity coefficient, which slows down the cooling of a microdroplet, can also play a certain role. In such case, a microdroplet does not have enough time to cool down and no secondary structures manage to form on its surface. Accordingly, when such a microdroplet (in liquid state) hits the deposition surface, it activates the mechanisms described in [14–19].

**Figure 5.** Examples of the influence of microdroplets on the structure of (**a**,**b**) Ti–TiN–(Ti,Nb,Zr,Al)N and (**c**,**d**) Ti–TiN–(Ti,Cr,Mo,Al)N coatings (**a**,**c**,**d**—SEM 1,**b**—SEM 2).
